Release history
Changelog
Every improvement to Geo Scorely, in one place.
Platform pivot: 6 focused hubs replacing 28 features — Score, Fix, Earn
🎛️ Cockpit — 10-dimension score gauge + top 5 prioritized fixes + live stats on one page
🤖 Autopilot — one-click setup: crawls 12 pages, extracts brand DNA, queries AI live, projects score, configures everything in 90s
👂 What AI Says — merged Live Mention Feed + Hallucination Detection into one hub
🚨 Hallucination CORRECTION — auto-drafts Wikipedia/Wikidata/Crunchbase/G2/LinkedIn patches for each detected hallucination
⚡ Fixes — merged AI Fixer + Content Gap Mining + Prompt Discovery into one action hub
📝 Auto-Brief Generator — Claude writes ready-to-use content briefs from gap analysis (titles, sections, entities, FAQ, schema)
💰 Revenue Attribution — full funnel: AI → visitor → lead → $. 4 attribution models. Per-engine breakdown
🧩 Install — free widget hub: Schema Injector + Crawler Firewall + Revenue Tracker, all free-to-install with usage caps
📊 track.js rewritten — AI referrer detection, localStorage event queue, 402 quota enforcement
🏢 White-label config for Agency/Scale tiers (logo, colors, custom domain, DKIM)
Free-install widget model: schema.js + firewall.js + track.js install for $0, monetized by usage limits
7-tier pricing: Free → Scan $29 → Starter $49 → Pro $99 → Growth $199 → Agency $499 → Scale $999
Left sidebar navigation with 6 hubs (replaces 14-item top-bar)
Autopilot reasoning traces — per-feature decisions with confidence scores + risk flags
Usage-limits helper with per-tier event caps, schema type limits, firewall retention
Fixed attribution horizontal privilege escalation (session_id scoped to user_id)
Fixed firewall fail-open on missing key (now returns 401)
Removed 18 feature pages, 14 lib files, 3 widget JS files for deleted features
Added 50 new GEO citation pattern rules (Tiers 11–23) including compound relationship detection
Dimension score insights — each dimension now explains why your score is high or low
"View fixes for this dimension" button filters advice panel directly
Compound rules detect multi-problem patterns: Answer-desert, Ghost site, Monolithic wall, and 7 more
Advice panel now returns up to 20 items (was 12)
Query-intent matching rules for "best", "how to", "vs", and "price" pages
Multi-strategy fetch pipeline: 6 strategies across 3 phases for maximum crawl success
Wayback CDX API integration for URL discovery
Entity density scoring — tracks % of proper nouns/technical terms (target: 15-25%)
Structured content density — measures % of content in bullets/tables/FAQs (target: 40%+)
Answer latency score — measures distance from H2 to first direct answer
Robots.txt bot monitoring expanded to 15 AI bots
FAQ answer length scoring (too long >70 words, too short <20 words)
PRISM MERGE multi-LLM orchestration — more API keys = exponentially better advice
Perplexity stat decay detection — flags statistics older than 24 months (-65% citation penalty)
Claude intellectual honesty rule — detects absolute vs. hedged language (1.7× citation boost)
SpeakableSpecification schema detection
Wikidata sameAs link detection for Knowledge Panel alignment
Fixed JS-rendered site detection false positives on React SSR sites
10-dimension Signal Forge scoring engine (v2) based on Princeton GEO research
Content-Answer Fit dimension — measures how directly pages answer target queries
Competitor citation gap analysis
AI Rewrite panel (BYOK) — rewrite content for AI citability
Score calculation now uses variance-adaptive power mean algorithm
Complete scoring engine rewrite — from v1 heuristics to research-backed v2 model
Multi-page crawl support — analyze up to 50 pages per scan
Schema.org validation with error detection
Citation performance tracking across ChatGPT, Perplexity, Gemini, Claude, Bing
Team collaboration features with role-based access
Scheduled monitoring with email alerts